**3. Animal Models in Drug Discovery and Development against Trypanosomatids**

Animal models are expected to mimic the pathophysiological features and immunological responses observed in the human host. A good experimental model for parasitic infections allows estimation of the specificity of drug action in relation to absorption, distribution, metabolism, excretion, and toxicity. Experimental models like rodents, dogs, and monkeys have been developed in order to identify and profile novel drugs against trypanosomatids, though mimicking the pathogenesis of disease and the impact of natural transmission is difficult to emulate under laboratory conditions [69]. The genotypic feature of laboratory models also augments hindrances due to restricted genotypic variations compared to infection with wild varieties. Hence, animal models developed and practiced for *T. brucei*, *T. cruzi*, or *Leishmania* infections do not accurately reproduce the consequences in human hosts, though several of these models exhibit an acceptable degree of proficiency for drug and vaccine development, particularly for the in vivo testing of trial compounds and libraries [70]. Important among them are BALB/c mice and Syrian golden hamster (primary tests), dogs (secondary tests), and monkeys (tertiary screens) as models for VL alongside athymic and SCID mice, which serve as a model for the treatment of VL in immunosuppressed conditions [69,71]. The genetic basis of the degree of susceptibility of mice to *Leishmania* has been linked to the Sc11 1a1 locus, based on which the outcome can be either self-healing or fatal [72]. The widely used (BALB/c and C57BL/6) mice breeds are mutated in the locus. In BALB/c mice, the immunopathology does not actually resemble human infection; instead, after around four weeks of infection, a strong Th1 response results in clearance of the parasite from the liver [72]. BALB/c is also highly susceptible to infection by *L. major,* with severe lesions and parasite-specific Th2 response with the enhanced expression of deactivating macrophage cytokines—particularly interleukin 4 (IL-4), interleukin 10 (IL-10), and transforming growth factor-β (TGF-β) [73]. On the contrary, the majority of inbred mouse strains like CBA and C57BL/6 are resistant to infection by *L. major,* and lesions spontaneously heal in 10–12 weeks [73]. The situation is bit different for the new-world *L. mexicana* and *L. amazonensis*, for which BALB/c, C57BL/6, and CBA/J mice are susceptible to infection [70]. On the contrary, for *L. braziliensis*, majority of mouse strains are resistant as the parasite does not induce protective Th2 response in the host [74]. However, for BALB/c, co-administration with salivary gland exudates of the vector promotes infection by altering the cytokine milieu [74]. Genetic susceptibility studies identified that the scl-1 locus controls the healing versus non-healing responses to *L. major* and the scl-2 is ascribed to the development of *L. mexicana*-induced cutaneous lesions. Around 30 loci have been identified as involved in the complex control of cutaneous leishmaniasis (CL) in mice [75]. BALB/c mice have been exploited as a model to profile metabolic changes during infection by *T. brucei* [72]. Mouse models including BALB/c, SCID, C57BL/6, and CH3 are the most widely used animal models in Chagas disease research [76]. However, the outcome was different in terms of Chagasic cardiomyopathy based on the strain of parasite and mouse line chosen for infection. Among alternative rodent models, guinea pigs have also been used as a model for experimental *T. cruzi* infection for acute and chronic Chagas disease [77–79]. For *T. brucei*, Wistar rats have been exploited as a preclinical model for HAT-associated cardiomyopathy [80]. The cotton rat (*Sigmodon hispidus*) represents one of the most susceptible animal hosts for *L. donovani*. The infection remains for 3–4 months, and after the appearance of initial clinical signs, the disease progresses rapidly, leading to death of the host [81]. Among various hamster species that are susceptible to *L. donovani*, the Syrian golden hamster (*Mesocricetus auratus*) represents a good model for VL with synchronous infection in the liver and spleen that culminates into a chronic non-cure infection with immune responses similar to human VL [81]. However, optimization of this model for drug screening is also effectively achieved through an ex-vivo splenic explant [45]. The only model that shows true potential for the evaluation of potential drugs targeting *L. braziliensis*, with low virulence for mice, is the golden hamster. Disease progression can be monitored over longer periods due to the chronic nature of the disease in the hamster [82]. For *L. infantum*, dogs are the natural reservoir. The natural infection of domestic dogs with *L. braziliensis*, *L. panamensis* and *L. mexicana* has been reported in Latin America. The infection of dogs with *L. infantum* is a pertinent laboratory model because it reproduces the natural

infection with considerable similarity to human infections. The use of dogs as experimental models to study VL actually elucidated the role of immune cells, cytokines, and signaling events mediating immune response during *Leishmania* infection, offering crucial clues for developing immunotherapy. Canine models of *L. mexicana* infection have been established with Beagle dogs [83].

Non-human primates are exploited as the first experimental model for evaluating safety and efficacy of drugs and vaccines. For VL, *Macaca* sp. developed low and/or inconsistent infections. However, *Presbytis entellus* showed substantial susceptibility to hamster-derived amastigotes of *L. donovani* with all the clinical-immunopathological features as observed in kala-azar characterized by consistent and progressive acute fatal infection, leading to death between 110 to 150 days post-infection. The *L. major*–rhesus monkey model emulates self-limiting human cutaneous leishmaniasis that resolves within three months [73,84,85]. The model also shows promise in deciphering the intricacies of immune function and granuloma formation by *L. braziliensis*, rendering it as a useful model for drug and vaccine development [86]. Non-human primates have been explored as models for Chagas disease, but in most of the studied cases only a limited number of animals develop typical cardiomyopathy signifying *T. cruzi* infection [87]. Recent analysis of circulating leukocytes from naturally infected non-human primate cynomolgus macaque revealed a strong resemblance with immune-pathological biomarkers of Chagas disease in humans, projecting the prospect of this model in preclinical studies for new drugs for Chagas disease [87].

#### **4. Cheminformatics in Drug Discovery**

After the identification of several important and prospective drug targets like reductases of folate metabolic cascade, kinases, cAMP-phosphodiesterases, and enzymes for trypanothione synthesis and purine salvage, cheminformatics studies to identify structure–activity relationships for the design of optimized compounds have been prioritized. In recent times, combinatorial chemistry and HTS have enabled tests on large compound libraries, which encompass a significant chemical diversity, in short time scales [88,89]. Cheminformatics tools are broadly classified into structure- and ligand-based drug design (SBDD and LBDD) approaches. SBDD exploits the 3D coordinates of target structures for favorable ligand interactions. Potential ligands can be screened by molecular docking or structure-based virtual screening of potential ligands. High-affinity interactions between the binding site and ligand can be achieved by exploring binding site attributes like electronic distribution. The establishment of structure–activity relationships (SARs) can be achieved through experiments to further optimize ligand–receptor affinity [90]. Alternatively, ligand-based drug design studies can be performed without the receptor 3D structure. Instead, they require information on the structure, activity, and molecular properties of small molecules [91]. Chemometric models based on quantitative structure–activity and structure–property relationships (QSAR and QSPR, respectively) can be built in order to identify molecular descriptors complementing the target property [92].

Pteridine reductase (PTR1), an enzyme of the folate biosynthetic pathway, was one of the prominent candidates for drug targeting since no homologue of that protein is detectable in mammalian hosts. The crystal structure of LmjPTR1 was determined [93]. Implementing an SBDD strategy, Rasid et al. (2016) identified a number of dihydropyrimidine- and chalcone-based inhibitors for *Leishmania* PTR1 [94]. Using homology model for type 2 NADH dehydrogenase, Stevanovic et al. (2018) conducted a pharmacophore-based virtual screening to identify several hits [95]. A 6-methoxy-quinalidine derivative showed potential inhibition of the recombinant protein and inhibition of amastigotes with an EC<sup>50</sup> of nanomolar range. Tryparedoxin peroxidase, a parasite-specific enzyme and a key component for parasitic survival under macrophage oxidative stress, has been considered as a key drug target. By performing deep molecular docking analysis with the crystal structure of PTR1 from *L. major*, a series of *N,N*-disubstituted 3-aminomethyl quinolones was identified which might serve as a worthy starting point for a suitable drug. SAR analysis of benzimidazole inhibitors against cysteine proteases cruzain and rhodesain from *T. brucei* and *T. cruzi*, followed by detailed cheminformatic analysis was conducted to find scaffold novelty and favorable physicochemical properties. Distinct endopeptidases like

cathepsin-L-like CPB2.8 have emerged as exploitable drug targets in leishmaniasis. De Luca et al. (2018) identified a group of substituted benzimidazole derivatives that displayed strong (nanomolar) affinity for the protease from *L. mexicana* [96]. One of the compounds demonstrated a good bioavailability profile with ADMET analysis, implying it is a good future drug candidate. Carbonic anhydrases (CAs) have recently been identified from trypanosomatids. Cheminformatics analysis targeting this enzyme identified *N*-nitrosulfonamides as prospective inhibitors for CA from *Trypanosoma* and *Leishmania* over mammalian homologues. Being comparable with existing drugs in terms of EC<sup>50</sup> and cytotoxicity, these compounds might serve as interesting leads for drug development.

Using the ligand-based approach, aminophosphonates have been studied with QSAR modelling [97]. The authors took the gathered data for the whole compound series to build comparative molecular field analysis (CoMFA) models that suggested that several modifications can enhance the anti-leishmanial potential of α–aminophosphonates. Similar approaches identified 1,2,3-triazole and thiosemicarbazone hybrids and tetrahydro-β carboline derivatives as candidate anti-leishmanial drugs [98]. Novel quinazoline and arylimidamide derivatives have been identified using 3D QSAR-based analysis against *T. cruzi* [99]. The structure-guided discovery of a compound (compound 7) from the pyrazolopyrimidine series against a known protein kinase scaffold identified *Leishmania* CDK12 as a strong candidate for drug discovery. Structural studies combined to resistance mechanism analysis confirmed CDK12 as a specific target for the molecule [99]. With satisfactory specificity as well as pharmacokinetic and toxicological properties, the compound has been declared a preclinical candidate, suggesting cheminformatics can indeed boost systematic approaches to discover new drugs against trypanosomatids [99].

#### **5. Quiescence, a Double-Edged Sword in the Quest of New Trypanocidal Drugs**

Dormancy or persister cell formation is an evolutionarily conserved adaptive mechanism for stress tolerance for bacterial pathogens. Persister cell development is often associated with the development of a subset of a population that is metabolically quiescent and hence cannot be intervened by drug treatment [100]. Such an adaptation enables the parasite to survive under immunological stress and drug exposure, reverting to normal proliferative mode once the stresses disappear. Such conditions are well exemplified by the latent infection of *Mycobacterium tuberculosis* which can persist for the entire lifespan in a metabolically dormant state [101]. Similar metabolic diversions from proliferative to dormant state are observed in eukaryotic pathogens including fungal and parasitic protozoan infections [102]. The hypnozoite liver stages of *Plasmodium*, often associated with relapse of infection even years after successful therapeutic clearance, is one such persister-like stage for *Plasmodium vivax* [103]. For trypanosomatids, semi-quiescence to quiescence have been detected for intracellular forms of several species of *Leishmania* and in *T. cruzi* [102]. Persister formation is particularly relevant clinically for *Leishmania*, as relapsing conditions like post-kala-azar dermal leishmaniasis (PKDL) occurring several years after treatment for visceral leishmaniasis and leishmaniasis recidivans occurring after the treatment of cutaneous leishmaniasis emerge from possible metabolically distinct parasites that circumvent drug treatment due to dormancy without acquiring resistance by signature genetic alterations [104]. Despite its clinical significance, there has been a lack of concerted effort to study persister development in trypanosomatids due to technical constraints including the labelling of quiescent cells to distinguish them from the normally proliferating population. In 2015, a detailed identification and characterization of the semi-quiescent physiological state was reported in *L. mexicana* intracellular amastigotes in infected BALB/c non-healing lesions with a prolific increase in doubling time to ~12 days compared to ~4 days in ex-vivo macrophage infections [105]. The semi-quiescent metabolic state was also characterized by low rates of transcription and protein turnover that is distinct from stationary phase or metacyclic promastigotes, and is possibly a response to complex growth restriction in the intracellular microenvironment in granulomas. They identified two distinct macrophage populations, one with ~100 cells and the other with an average of ~400 intracellular amastigotes, suggesting the existence of two distinct metabolic amastigote varieties. *L. mexicana* amastigotes are

intrinsically more resistant to nitric oxide and build up large communal phagolysosomes, while *L. major* infection is eventually controlled by an adaptive Th1 immune response requiring inducible NOS (iNOS) [105]. Mandell et al. (2015) identified a definite fraction of amastigotes with barely detectable replication in a C57BL/6J mouse model of cutaneous *L. major* infection. This population was observed to harbor in less-infected macrophages and constituted almost 39% of amastigotes under the persistent infection condition, while a second subset of amastigotes retained the ability to replicate with a doubling time of around 60 h [106]. *L. major* lacking the Golgi GDP-mannose transporter required for lipophosphoglycan synthesis encoded by LPG2 (lpg2-) persist in the absence of pathology, and in mouse infections this knocked-out line attained a persister-like feature immediately after infection [106]. *L. braziliensis* amastigotes (both axenic and intracellular) bear characteristic features of quiescence, with a radical reduction of (i) the kDNA mini-circle abundance, (ii) the intracellular ATP level, (iii) the ribosomal components, and (iv) total RNA and protein levels [107]. The untargeted metabolomic profile revealed the significant depletion of amino acids, polyamines, and trypanothione, with increases in ergosterol and cholesterol biosynthesis. Dormancy attains further relevance for trypanosomatid infection, as regimens including short-term therapy of even 60 days for *T. cruzi* infection is not related to resistance development, and the parasite possibly alleviates drug-mediated clearance by adopting quiescence. In fact, in *T. cruzi*, non-proliferating amastigotes develop both in vitro and in vivo models of infection. *T. cruzi* amastigotes regularly and spontaneously cease replication and become non-responsive to effective trypanocidal drugs like benznidazole and nifurtimox [108]. One or two such dormant parasites are detectable in each infected cell after treatment. Such dormant parasites reinitiate proliferation after drug withdrawal. Exploring the intricacies of the alteration of physiological status for intracellular amastigotes in infected tissues by proteomic or transcriptomic approaches is impaired by the paucity of enrichment protocols. Each of these studies adopted various strategies to characterize and label persister cells. One such strategy exploited <sup>2</sup>H2O labelling for determining DNA, RNA, protein, and membrane lipids. The in vitro deuterium labelling of deoxyribose could be achieved for promastigotes by maintaining 5% <sup>2</sup>H2O in medium, and for the in vivo labelling of amastigotes, 5% <sup>2</sup>H2O in the body water was established by providing mice with a bolus of 100% <sup>2</sup>H2O followed by inclusion of 9% <sup>2</sup>H2O in the drinking water for up to several months [105]. Differential labelling for replicative and non-replicative amastigotes is achieved with CellTrace Violet or CellTracker Red. After a brief pulse, the stain is either diluted out during cell division (for replicative form) or remains at the initial pulse level (for non-replicating forms). This approach can be combined with a fluorescent (tdTomato) or luciferase expression system to track viable parasites [108]. The incorporation of thymidine analogues 5-ethynyl-2′ -deoxyuridine and 5-bromo-2′ -deoxyuridine has been implemented to differentiate replicative and non-replicative cells in *Leishmania* spp. and *T. cruzi* [108,109]. Each of these approaches has been effective in tracing persister cells. Active translation or ribosomal action utilizes 70% of the total ATP generated in a viable cell, and in quiescent cells translational activity is highly compromised, with a concomitant decrease in the number of active ribosomes (~5-fold reduction in dormant compared to normal metabolic state). Hence, the reduced transcription of rDNA loci serves as a marker for quiescence and rDNA loci are part of a rare genomic landscape in trypanosomatids, which is regulated by a definite transcription factor [110]. In this context, the expression of the GFP gene under the 18S ribosomal DNA locus has been implemented as a biosensor for quiescence in laboratory and clinical strains of *L. braziliensis* and *L. mexicana*, and reduction of GFP expression was compatible with BrdU uptake analysis in vitro. With this approach, a superior FACS quantitative approach for persisters could be devised for recording quiescence development in mice (BALB/c) or hamsters (LVG Golden Syrian Hamster) models [109]. The study provided a clearer idea about metabolic diversity in amastigotes with the coexistence of shallow and deep quiescent stages. Quiescence is crucial for subclinical infections with its potential role in drug tolerance, and quiescent cells serve as reservoirs for transmission and elicit a protective response against subsequent infections in trypanosomatids, which warrants additional exploration [106]. The development of novel assay methods combined with

identification of strategies to combat dormancy or exploit it in developing immunization strategies might expedite the success of elimination programs against trypanosomatid parasites.
